The task
Lead follow-up means deciding each day which leads are due for a nudge, checking whether they have already replied, and writing a short message in the sender's own voice. The source is a tutorial on automating that routine for a lead list kept in Airtable. The author reduces the job to three plain steps: read the leads, decide who is due, and draft the message.
What they did
The agent is built with the Claude Agent SDK and reaches Airtable, Gmail, and Slack through MCP servers. According to the source's example, a system prompt sets the rules: create drafts only, never send directly, contact a lead no more than once in 7 days, stop the sequence when a lead replies, and log every action. A separate markdown skill file lists the daily steps. The agent fetches the leads that are due, checks Gmail for a reply from each address, marks repliers as active, and writes a Gmail draft for everyone else. According to the source's example configuration, a scheduled task starts the run on weekdays at 9. The drafts are then posted to Slack for review.
What the source reports
According to the source, one daily run on 50 leads uses roughly 30-80k tokens with Sonnet, which the author puts at under $1 per day per agent. The source gives no figures for replies, meetings, or revenue.
Where the approval sits
The agent creates drafts and never sends. It posts them to Slack, and the builder approves or rejects each one. According to the source, the author runs every outbound agent in this approval mode for at least two weeks per customer segment, and keeps it for sales outreach until the draft quality is consistent.
Worth knowing
The source is a tutorial by an individual automation builder, with example prompts and code rather than a report from a client project. The cost figures are the author's estimate. The post names three ways the setup breaks: outdated skill files, missing tool permissions, and an approval queue that nobody reviews. It also describes a mode that sends drafts automatically and points to the author's own community for further material. This entry leaves both out and describes only the reviewed setup.
Source: Developer community post · Apr 2026
Drafted with AI assistance and checked automatically. A person is responsible for the collection. How these entries are made
Prompt for your agent
Draft check-in notes for quiet leads
Turn an export of your lead list into short check-in notes for leads that have gone quiet, each ready for your review.
More in this category
- Sales follow-ups
A sales team has an agent draft post-call emails, and reps edit and send them
A sales team uses an AI agent to draft the follow-up email after each sales call, based on the call summary. The rep edits every draft and sends it personally.
Read the entry - Sales follow-ups
A weekly agent drafts outreach to lookalike accounts and posts it to Slack for review
A company co-founder set up a weekly Claude Cowork task that turns recent closed-won deals into draft outreach for similar companies. The drafts are posted to Slack where the team reviews them, and sending is not automated.
Read the entry
Ninety minutes live on your tools. One automation running before we hang up.